BearTec AI · a BEARTEC LTD service

Private AI that runs in your office.

We install a local AI “Brain” on your premises, set it up around how your business actually works, and build the tools your team uses every day: assistants, searchable knowledge, engineering models and job systems.

Built and proven first inside a family-run industrial combustion engineering business, taking it from paper to an AI-assisted operation.

  • Runs on your own hardware
  • Drafts only, a person approves
  • Fixed-price packages
Scroll

01 · Flagship package

The BearTec Brain: your own AI box, set up for you.

A workstation-class machine that lives in your office and runs open-weight AI models locally. We build it, load it with an agreed set of your documents, set up the workflows you care about, install it on site and show your team how to use it.

  • Hardware supplied and bench-built (about £3,000 of the package at today’s prices; exact spec confirmed at quote)
  • Local model runtime and chosen open-weight models
  • Your documents ingested so it can search and answer from them
  • Workflows set up: drafting quotes, summarising job sheets, searching history
  • On-site install, network setup and staff handover
  • 30 days of snagging, then an optional monthly care plan
BearTec Brain package from £9,500 Indicative, fixed-price, hardware included. Final quote after a discovery call.
Ask about the Brain
Brain console showing the local models installed and loaded
Phone: real Brain console screen, run on demo data.

02 · In-house coding agents

Small tools, built by AI, signed off by you.

The same Brain can run coding agents that build and maintain the little internal tools every business needs: stock trackers, reports, dashboards. Code and data stay on your box.

  • Agents work on a branch and open a draft change for review
  • Nothing goes live until a named person approves it
  • Version control and backups set up from day one
  • One or two of your staff trained to direct and review the work
brain.local / console
Left: real Brain console on demo data. Right: illustrative workflow.

03 · Automated computer modelling

From parameters to tested designs, automatically.

Our first build models a natural-gas fan burner head. Parametric CAD generates the geometry, CFD cases are built and run without hand set-up, and an optimiser searches for the best head design for each power rating.

  • Parametric CAD (CadQuery) producing CAD and STL files
  • Automated OpenFOAM cold-flow CFD runs on local hardware
  • Multi-objective optimisation (Optuna NSGA-II) with results summaries
  • Scripts and a run guide handed over to your team

Outputs are design aids. A competent engineer reviews and signs off every design.

burner-rnd · showcase
CFD gas mixing cross-section at 120 kW
CFD gas mixing @ 120 kW
Live solver convergence dashboard
Live solver convergence

04 · Custom assistants & avatars

An assistant on every desk that knows how you work.

Spark sits in the corner of the screen. Staff click it to plan a quote or a job step by step. It tidies rough typing into proper wording, picks up each person’s preferred tone, and drafts the scope ready to paste.

  • Answers from your documents, customer lists, parts and job types
  • Each person sets their own style once: tone, formality, sign-off
  • Draft-only: nothing is sent or saved without a person confirming
  • Choice of characters, always presented as an AI assistant
Desktop
Spark first-run setup: pick a character and writing preferences Spark drafting a works scope from a rough request
Real Spark overlay, run locally on demo data.

05 · AI knowledge databases

Every job sheet you’ve ever written, finally searchable.

We pull in your documents, tag them with a local model and build a linked knowledge graph and Obsidian vault: customers, sites, equipment, jobs, quotes, invoices and parts, all connected.

  • Document intake and cleaning (for example from OneDrive)
  • AI tagging on your own Brain, with claims checked against the source text
  • Drill down by customer, site and plant in a 3D web or plain notes
  • A search aid for your team, never an official classification
knowledge web
Real 3D knowledge web, rendered with invented demo records.

06 · CRM, inventory & on-site tools

Off paper, onto one system your engineers actually use.

We build the operations suite around your process, module by module: customers and enquiries, quotes, jobs, invoices, stock and suppliers, rotas and timesheets, plus a mobile app for engineers on site.

  • Discovery workshop and process map first
  • Engineer app with on-site calculators that produce signed certificates
  • Job sheets, test records, RAMS and photos in one place
  • AI helpers that draft write-ups for the engineer to accept
  • Imports from your spreadsheets and old service history
Gas tightness test and purge certificate generated by the engineer app, BearTec branded, demo data
Engineer app tools list: project builder, test and purge, ventilation and pipe sizing calculators
Engineer app tools list and a test & purge certificate it produces, from our own build, shown with invented demo data.

How it works

From first chat to a Brain your team relies on.

  1. 1

    Discover

    We map how your business runs, which documents matter, and the jobs you want AI to help with. You get a fixed-price scope.

    Week 0
  2. 2

    Install the Brain

    We bench-build the machine, load the models, then install it on your network and check it end to end.

    About 1 to 2 weeks
  3. 3

    Tailor

    Your documents go in, workflows are set up, and each person’s preferences are captured so drafts sound like your business.

    On site, 1 to 2 days
  4. 4

    Handover & support

    Staff sessions, a run guide, 30 days of snagging, then an optional care plan for updates and new workflows.

    30 days, then optional

Typical Brain package timeline is 2 to 4 weeks from deposit. Hardware lead times can vary.

Questions

The things people ask first.

Does our data leave the building?

The Brain runs open-weight models on hardware in your office, so your documents are processed locally. If a cloud model would help with a particular job we will say so, explain what would be sent, and it stays off unless you opt in.

Is the AI “trained” on our business?

We set it up around your business rather than retraining the model itself. Your documents are indexed so it can search and answer from them, and your workflows, templates and each person’s preferences shape how it drafts.

What hardware do we get?

A workstation-class machine of the same class we run ourselves, roughly £3,000 of the package at today’s prices. The exact specification is confirmed in your quote. You own it.

Can it send emails, quotes or invoices on its own?

No. Everything we build is draft-first. A person reviews and approves before anything is sent, saved or put live.

Do we need our own IT team?

No. We install it, set up the network side and train your staff. An optional monthly care plan covers model, driver and system updates, and new workflows.

Are the engineering models safe to rely on?

They are design aids. A competent engineer must review and sign off every design, and that is written into the scope.

How do we get a price?

Email us with a little about your business. We’ll arrange a discovery call and send a fixed-price scope. The Brain package starts from £9,500; other products are quoted to your scope.

Want private AI working in your office?

Tell us what your team spends its time on. We’ll tell you honestly what AI can and can’t do about it.

Email beartec@beartec.uk